| f7221b0cbaf597a0… | The CONTEXT_FOR edge connects a CxU to an entity with dimension, q_value, and is_rank0 properties so that contextual relevance and ranking metadata can be attached to knowledge units. | active | contextual | yes | |
| dbbdfca190393c9b… | The Neo4j graph schema models memory using MemoryNode entities connected by TEMPORAL_NEXT, SEMANTIC_SIMILAR, and REFERENCED relationships so that temporal, semantic, and citation-style memory links can be stored. | active | contextual | yes | |
| 5414bddd51b7df2f… | The Neo4j graph schema includes AgentRun, ActionItem, and Analysis as agent nodes so that agent activities and outputs can be represented as first-class graph entities. | active | contextual | yes | |
| 22d97169310d4b7c… | The Graph Schema in Neo4j organizes hierarchy nodes as Company, Pillar, Group, BusinessUnit, VirtualBusinessUnit, and CrossUnit so that organizational structure can be represented explicitly in the graph. | active | contextual | yes | |
| 2e6a4eb2a9b130cc… | The DraftCXU data model defines a CXU record with identifiers, a standalone claim, 3 to 6 supporting quotes, classification fields, assumption linkage, keywords, and a confidence score so that extracted knowledge can be stored in a structured and traceable format. | active | contextual | yes | |
| bcd82e49fc271272… | The security configuration applies seccomp syscall filtering on Tier 1 so that low-level system calls are restricted for subprocess-based executions. | active | contextual | yes | |
| 28697e79f8a4734d… | The security configuration sets a 10 second execution timeout that is configurable per plan so that executions have a default upper bound while allowing plan-based adjustment. | active | contextual | yes | |
| be62cf1ce2e57def… | The security configuration enforces a PID limit of 50 to prevent fork bombs so that runaway process creation is constrained by the runtime. | active | contextual | yes | |
| 257a70bcdecbe6a1… | The security configuration limits memory to 128MB and CPU to 0.5 cores so that code execution remains resource constrained within the interpreter environment. | active | contextual | yes | |
| 3be4680f0c431446… | The security configuration disables network access in containers so that code running in containerized environments cannot communicate externally over the network. | active | contextual | yes | |
| d7ebcb866ca249a4… | The GET /metrics API endpoint exposes concurrency metrics including active_executions, max_concurrent, and available_slots so that clients or operators can inspect execution capacity and usage. | active | contextual | yes | |
| eac3f09339232ebe… | The GET /health API endpoint is used for Kubernetes probes so that orchestration systems can verify service health. | active | contextual | yes | |
| 83f528dcda2937e2… | The POST /run API endpoint executes Python code and can optionally accept workspace, input_files, and output_files parameters for Azure Blob I/O so that code execution can interact with external file storage. | active | contextual | yes | |
| 922c49370d99a9dd… | The PoolStateManagerAgent consumes all events and auto-scales every 30 seconds so that pool capacity can be adjusted continuously based on system activity. | active | contextual | yes | |
| 91548022ec7c5728… | The QueueOptimizerAgent polls queue depths every 1 second and activates backpressure at 80 percent so that the system can respond quickly when queue load becomes high. | active | contextual | yes | |
| 87a30c052b563289… | The HealthMonitorAgent checks pool slot health every 10 seconds and restarts FAILED slots so that unhealthy execution slots are automatically recovered. | active | contextual | yes | |
| b138ba4f7511be8e… | The ResourceMonitorAgent polls /proc/meminfo every 5 seconds and tracks memory pressure so that the system can monitor resource usage continuously in the background. | active | contextual | yes | |
| e5f500c6423cf1cb… | The code classifier determines execution complexity as LIGHT, STANDARD, or HEAVY based on imports present, code size, timeout, file I/O, and output file count so that code can be routed to an appropriate execution tier. | active | contextual | yes | |
| 113ea818d43dc5a5… | The Silver subscription plan includes Tiers 1 and 2, allows file I/O and data-science libraries, disallows ML or GPU libraries, permits 20 concurrent executions, and sets a maximum timeout of 120 seconds. | active | contextual | yes | |
| 31eb7ac90ecbf455… | The Bronze subscription plan includes Tier 1 only, disallows file I/O, data-science libraries, and ML or GPU libraries, permits 5 concurrent executions, and sets a maximum timeout of 30 seconds. | active | contextual | yes | |
| 6e391e930f9bb68d… | Tier 3 uses on-demand Firecracker microVMs with under 2 second P99 latency and VM-level isolation for ML training, GPU workloads, and system dependencies so that the heaviest workloads run in the strongest isolation tier. | active | contextual | yes | |
| 2cf2a13816fba21a… | Tier 2 uses pre-warmed Docker containers based on python:3.11-slim, a pool size of 5 with a 2-10 range, under 500 millisecond P99 latency, and container-level isolation for pandas, numpy, CSV or Parquet I/O, and data pipelines. | active | contextual | yes | |
| 7fa58db5e334cd65… | Tier 1 uses forked Python subprocesses with seccomp, a pool size of 20 with a 5-40 range, under 100 millisecond P99 latency, and process-level isolation for arithmetic, strings, and simple algorithms. | active | contextual | yes | |
| 6ee3ba9dd3711153… | The tiered code interpreter architecture replaces a legacy single-tier Firecracker VM approach with three execution tiers mapped to subscription plans so that execution environments align with customer plan levels. | active | contextual | yes | |
| 893bb83e61863fa8… | The code interpreter tiered architecture executes Python code in isolated environments for AI agents to generate structured JSON data programmatically so that agent-driven computation can run safely and produce machine-readable outputs. | active | contextual | yes | |
| 518da7712e5e534d… | The CORTEX_LEARNING_ENABLED flag enables Q-value learning so that the platform can activate learning behavior tied to Q-value updates when desired. | active | contextual | yes | |
| 7cde7d9fc8cb188c… | The CORTEX_AMBIENT_ENABLED flag enables the ambient research cycle so that background research behavior can be activated independently of other CORTEX features. | active | contextual | yes | |
| 5e2898f85a729a1e… | The CORTEX_RETRIEVAL_INCLUDE_DYNAMIC_RESULTS flag enables Strategy 3, identified as Text2Cypher, so that retrieval can include dynamically generated graph-query results when the flag is turned on. | active | contextual | yes | |
| 26dd3a0684546ebf… | The CORTEX_ENABLED flag is the master switch for CORTEX functionality so that the platform can enable or disable the overall feature set centrally. | active | contextual | yes | |
| d0696a0f7c32b9a6… | All CORTEX feature flags default to off so that advanced retrieval, ambient research, and learning capabilities remain disabled unless explicitly enabled. | active | contextual | yes | |
| 308fef8135943375… | The Qdrant payload filters are set_names, related_node_ids, and is_rank0_set_names so that vector queries can be constrained by set membership, related nodes, and rank-0 set status. | active | contextual | yes | |
| 53cdc70334b86cab… | The Qdrant vector storage defines collections named docs_{region}, user_memory_{region}, and session_memory_{region} so that data can be partitioned by both content type and region. | active | contextual | yes | |
| 985121ad4339c098… | The vector storage layer uses Qdrant with the all-mpnet-base-v2 embedding model at 768 dimensions so that document and memory vectors are stored in a consistent embedding space. | active | contextual | yes | |
| 37da47f5886325fc… | The IN_SET edge in the Neo4j schema links a CxU to set membership so that context units can be associated with named collections or sets. | active | contextual | yes | |
| 95034fd9207e324b… | The DERIVED_FROM edge in the Neo4j schema links one CxU to another CxU as lineage so that derivation relationships between context units can be tracked. | active | contextual | yes | |
| 667a1ac5db6227c9… | The PRODUCED edge in the Neo4j schema links an AgentRun to a CxU so that the system can record which agent execution generated a given context unit. | active | contextual | yes | |
| 43c6402d9a7492d9… | The BELONGS_TO edge in the Neo4j schema connects a child node to its parent within the hierarchy so that organizational parent-child relationships are explicitly modeled. | active | contextual | yes | |
| c53a41d371484e0d… | The CONTEXT_FOR edge in the Neo4j schema links a CxU to an entity and carries dimension, q_value, and is_rank0 properties so that contextual relevance can be represented with metadata. | active | contextual | yes | |
| 00b8c08d0a72abe8… | The Neo4j graph schema defines memory nodes as MemoryNode with TEMPORAL_NEXT, SEMANTIC_SIMILAR, and REFERENCED relationships so that memory structures can capture temporal, semantic, and citation-style connections. | active | contextual | yes | |
| ff35608bc588fb72… | The Neo4j graph schema defines agent nodes as AgentRun, ActionItem, and Analysis so that agent-related activities and outputs can be represented explicitly in the graph. | active | contextual | yes | |
| 0411f98f2079977d… | The Neo4j graph schema defines hierarchy nodes as Company, Pillar, Group, BusinessUnit, VirtualBusinessUnit, and CrossUnit so that organizational structure can be represented as a typed hierarchy. | active | contextual | yes | |
| e69cb031d2f77c15… | The DraftCXU model includes a keywords list and a confidence float ranging from 0.0 to 1.0 so that each CXU can support retrieval and express extraction certainty. | active | contextual | yes | |
| ec06b3186af3dd26… | The DraftCXU model includes is_assumption and parent_assumption_ids fields so that a CXU can indicate whether it is an assumption and record its assumption lineage. | active | contextual | yes | |
| d0535e94d58ac1ea… | The claim_type field in the DraftCXU model is defined as an enum used to classify the form of the claim, such as definition, requirement, or procedure, so that extracted units can be typed by expression style. | active | contextual | yes | |
| 7aa4d5fc8ca551bb… | The knowledge_type field in the DraftCXU model is defined as an enum with values axiom, derived, and prescribed so that each CXU can be categorized by its knowledge status. | active | contextual | yes | |
| 66f5e488d87cec0b… | The supporting_quotes field in the DraftCXU model is defined as a list containing 3 to 6 verbatim source quotes so that each CXU preserves direct textual evidence from the source material. | active | contextual | yes | |
| 9e4c00813451f89d… | The claim field in the DraftCXU model is defined as a standalone assertion between 15 and 70 words so that each extracted unit remains self-contained and concise. | active | contextual | yes | |
| 41d84da6911365b2… | The cxu_id field in the DraftCXU model is defined as a string composed of the prefix "1220" followed by a SHA-256 hexadecimal value, producing a 68-character identifier so that each CXU has a standardized unique ID format. | active | contextual | yes | |
| 15cced43efe50c8e… | The DraftCXU data model defines a CXU record with fields for identifier, standalone claim text, supporting quotes, knowledge and claim classifications, assumption tracking, keywords, and a confidence score so that extracted knowledge can be stored in a structured and evaluable form. | active | contextual | yes | |
| d1c958daf6ccb96d… | The security model applies seccomp syscall filtering on Tier 1 so that subprocess-based execution is restricted at the system call level. | active | contextual | yes | |